The Internet Protocol

Contents Understand the role of the Internet Protocol (IP) Examine IP address classes Use Address Resolution Protocols (ARPs) Decode IP packet structure Examine IP on various physical networks

The Internet Protocol

Internetworking Concepts Different networks connected together

Physical-Transport Independence

IP layer Provides powerful logical abstraction Hides Physical Layer dependency Upper layer processes see a logical IP network

Functions of Internet Protocol Internet Protocol (IP) provides Datagram service Phisical network independence for higher layer processing Logical address for computers on network Independence from maximum transmission unit size Fragmentation and reassembly control These topics are examined in the next several viewgraphs

IP Datagram Service Ip makes use of “best efforts” service Similar to postal services Advantages Simplicity and less overhead Upper layers can build more reliable service Adequate for many networks -- LANs, frame relay

Need for Uniform Addresses (Logical Addresses) Phisical networks use different addressing schemes Ethernet networks use 6-byte addresses X.25 networks use 14-digit decimal codes ARCNET networks use 1-byte addresses How should nodes on a logical network be identified? Solution: Use a logical address to provide a uniform way of addressing all network nodes rregardless of their physical network connections

Message Size Limitations Physical networks support different maximum frame size Example: 1518 bytes for Etherne, 512 bytes for ARCNET, etc. Upper layers (TCP) del with message size of arbitrary length Problem: How do you send arbitrarily long messages to networks with packet-size constraints? A solution: datagrams, fragmentation, and reassembly

Names and Addresses LAN Addressing NIC addresses (like National Insurance Number) Broadcast technology No geography Network Addressing IP addresses (telephone numbers) Permits subnetting (like county and area codes) Gives routing capability Maps to NIC address through ARP Human Friendly Names Shows affiliations (like a normal mail address) Structured independently of IP Used to identify people, computers, networks, organistations. Maps to IP through DNS

Typical MTU Size Network Typical frame size (bytes) MTU (bytes) Maximum frame size (bytes) Ethernet 1024 1500 1518 IEEE 802.5 (4 Mbps) 4464 4508 (16 Mbps)* 1024/4096 17,756 17,800 ARCNET 508 512 X.25 128 4080 4096 *Assuming a token-holding time of 9 ms. MTU = maximum transmission unit

Fragmentation and Reassembly Control Many applications on hosts tend to use large message size File transfer, graphic applications Many wide area networks prefer smaller packet size in comparison with some applications Better use of buffer memory Smaller probability of error for each packet Smaller delay for priority packets Larger packet size in networks (such as LANs) can result in more efficient data transfer

Freagmentation and Reassembly Control (continued) What would happen to packets going from Host B to Host A? Note: TCP messages can be fragmented by sender

The Internet Protocol

Calculating an Address Class (continued) First Decimal Number in Dotted Decimal Address Minimum Maximum A 1 126 B 128 191 C 192 223 D 224 239 E 240 247

Range of Assignable Addresses netid hostid Class Minimum Maximum A 1 126. 0.0.1 255.255.254 B 128.0 191.255. 0.1 255.254 C 192.0.0 223.255.255 .1 254 D 224. 239. N/A N/A = not applicable Why is address 127.x.x.x not assigned?

Software Loopback Local machne can be addressed by 127.x.x.x “x” can be any value -- Typically, 127.0.0.1 is used for local host Also referred to software loopback test -- Packets never transmitted -- Packets copied from transmit buffer to receive buffer

Hostid and Broadcast Addresses Hostid of 0 is never assigned to an individual host An internet address with hostid of 0 refers to the network itself -- Example: 144.19.0.0 -- Refers to class B network 144.19.0.0 Directed broadcast addresses By convention, broadcast addresses have all 1 s in hostid field -- Example: 144.19.255.255 Important exception is software derived from BSD 4.2 UNIX -- Uses all 0s broadcast Limited broadcast address Broadcast address of 255.255.255.255 IP packets with this address usually do not cross router boudary Not all TCP/IP implementations support it

The Internet Protocol

The Problem: Need for Address Resolution How does a host know about another host’s physical address? Hard code knowledge of physical addresses? Can you logical (IP address) to determine physical address? -- Address resolution protocols

Dynamic Address Resolution Protocol Mechanism Host A broadcasts ARP request on network containing B’s IP address All nodes receive ARP request, but only B responds because its IP address is included in the ARP request B replies to A, with B’s physical address Assumption: requires broadcast capability on network (i.e., Ethernet, Token Ring, etc.)

ARP Request/Response Packet Structure Numbers in ( ) represent bits

ARP Refinements: Caching In previous example Host A uses ARP reply to build a local cache -- Cache contains <IP addr., Physical addr.> pair Host B is likely to reply to A Use ARP request to store A’s <IP addr., Psysical addr.> in cache Other host extract A’s <IP addr., Physical addr> from ARP request Machines booting on netork announce their <IP addr., Physical addr.> Other machines cache this information Also used for duplicate IP address detection

IP Address for Disless Nodes Workstations store their IP addresses in local storage media How do diskless workstations store their IP addresses? A solution: use Reverse ARP (RARP) -- Keep <IP addr., Physical addr.? Bindings on RARP server -- Potential for simplifying IP adress administration?

RARP Operation RARP uses same packet structure as ARP Sender address – PA Destination address = broadcast Ethertype = 8035 hex RARP uses same packet structure as ARP

RARP Operation (continued) Send broadcasts RARP request SENDER HA ← Sender’s physical address TARGET HA ← Sender’s physical address RARP servers respond with OPERATION TYPE ← reply TARGET IP ← Answer (requester’s IP address) DATA LINK DA ← Requester’s physical address Lssues RARP request storms Primary and backup RARP servers

BOOTP BOOTP makes use of UDP/IP to obtain IP addresses and other information BOOTP does not provide clients with bootstrap image It provides the name of the boot image Boot image is transferred using Trivial File Transfer Protocol (TFTP) To forward BOOTP requests across routers, routers must be configured with rekay agents to foeward BOOTP packets

Troubleshooting Duplicate IP Addresses and ARP Tables Nodes on an IP network must have unique IP addresses Otherwise, ARP tables are initialized with incorrect <IP addr., Physical addr.> mappings Symptoms of bad ARP tables are -- Users unable to access TCP/IP hosts -- Workstations and servers crashing -- Intermittent problems with applications not working Common results of duplicate IP addresses are ARP table corruption at workstations ARP table corruption at servers

Duplicate IP Addresses at Workstation Step 1 Workstation initiates FTP session to server

Duplicate IP Addresses at Workstation (continued) Step 2 Second workstation with duplicate IP address initiates FTP session to server

Duplicate IP Addresses at Workstation (continued) If the server receives a TCP/IP connection request from a second workstation with a duplicate IP address, the TCP/IP software may Ignore the second request Overwrite the server ARP cache entry with hardware address from second workstation Get confused and crash In either of the above choices, one or both of the workstations with the duplicate IP address will have connection problems

Duplicate IP Addresses at the Server Workstation tries to connect to VAX at IP address 144.19.74.102 If the SUN server at duplicate IP address 144.19.74.102 returns an ARP reply faster than the VAX, the workstation connects to the SUN server instead of the VAX What happens if the SUN server and VAX server also act as routers?

ARP Display Utilities Resolving duplicate IP address problems can be a challenging task on large networks Keeping good records of IP address assignments and hardware addresses of devices can help Use utilities to display and fix ARP cache entries -- Most UNIX systems have the arp utility arp -a arp -d hostname arp -s hostname hardware_addr Display all ARP entries in table Delete an entry from ARP table Add a new entry in ARP table. Entry is not timed out!

The Internet Protocol

Hands-On Exercise 3.1: Address Resolution Your instructor will guide you to Hands-On Exercise 3.1 in the Exercise Manual

The Internet Protocol

Hands-On Exercise 3.2: Observing Effects of Duplicate IP Addresses Your instructor will guide you to Hands-On Exercise 3.2 in the Exercise Manual

The Internet Protocol

IP Packet Structure Background: This exercise is a guided tour on the structure of IP packets. It will be done concurrently with the lecture, which will explain the IP structure. You will use the packet trace that you saved in an earlier exercise for understanding the IP packet structure. Objectives: Examine the IP packet fields Understand the functionality of the IP protocol

IP Packet Structure (continued) Run LANWatch at your workstation. If you forgot how to run LANWatch, see page 16 in the Exercise Manual. Load the filt TELNET. TR1 that contains the TELNET packet trace youstored in an earlier exercise. Highlight one of the red packets that contain IP protocol information and display it in the detailed format. Follow the instructions given to you by the instructor.

IP Field: Version Version field Indicates format of IP header Declares version of protocol to which datagram belongs Allows development of new protocols while network is operational What is the version of the IP packet on your screen?

IP Field: Internet Header Length Measured in 32-bit words Required because IP header contains variable length options field What is the internet header length of the IP packet on your screen?___________________________ Does the IP packet have an options field? Yes No

IP Field: Type of Service (TOS) Informs networks on Quality Of Service (QOS) desired

IP Field: Type of Service (TOS) What is the bit pattern for TOS of IP packet on your screen? __________________________ What is the TOS value? _________________

IP Field: Total Length Total length Length of datagram (octets), including IP header and data portion Maximum datagram size is 65,535 octets All hosts must be prepared to receive datagrams of 576 octets 512 octets of data and 64 octets of protocol overhead What is the total length for the IP packet on your screen? _________________ octets?

IP Field: Identification Set uniquely for each datagram Used as an aid in assembling fragments of a datagram What is the identification value for the IP packet on your screen? ___________

IP Field: Identification (continued) Use cursor keys (↑, ↓) to examine identification field values of IP packets before and after this IP packet What is the identification value of the previous IP packet?______ What is the identification value of the next IP packet?________

IP Field: Flags What are the flag settings for the IP packet on your screen? DF flag =_______ MF flag = _______

IP Field: Fragment Offset Position of fragment's data relative to the beginning of data carried in original datagram Maximum of 8192 fragments per datagram Identification field is same for all fragments

IP Field: Fragment Offset What is the fragment offset for the IP packet on your screen?_________________________

IP Field: Time to Live Time to live Maximum time IP datagram can remain on internet When TTL = 0, IP datagram is destroyed (dropped) Decreased by time for IP header processing, but must be decreased by at least 1

What is the TTL field value for the IP datagram on your screen

IP Field: Protocol Protocol field Indicates which Upper Layer Protocol (ULP) is to receive data portion of IP datagram What is the protocol field value for the IP packet on your screen?

IP Field: Protocol (continued) Protocol field value Keyboard Description Reserved 1 ICMP Internet Control Message Protocol 6 TCP Transmission Control Protocol 8 EGP Exterior Gateway Protocol 9 IGP Any private Interior Gateway Protocol 11 NVP Network Voice Protocol 17 UDP User Datagram Protocol 22 XNS IDP Xerox Network System’s Internet Datagram Protocol 29 ISO TP4 ISO Transport Protocol class 4 89 OSPF Open shortest path first

IP Field: Header Checksum Covers only the IP header Add up 1's complement of each data item (16-bit) and then the 1 's complement of the sum Recomputed at every route because TTL field changes What is the header checksum field value of the IP packet?_______________________

IP Field: Source Address, Destination Address Source and destination addresses are divided in netid and hostid fields What are the source-address and destination-address fields of the IP packet on your screen? Source address: _______________ Destination address: ____________

IP Field: Options Options Officially defined options are -- Security, loose source routing -- Strict source routing, record route -- Stream ID, Internet timestamp Options are of two types

Are there any options defined in the IP packet on your screen

Bonus Pick an IP packet (that appeals to you!) within LANWatch and analyze it on your own. Try to identify the blank fields in the IP datagram above. Label these fields and enter the value in the fields for the IP packet you are analyzing.

The Internet Protocol

Duplicate IP Address Problem IP addresses must be unique Most network software assumes trusted hosts Duplicate IP addresses result in Network software becoming confused, malfunctioning Routing problems -- Because routing information is encoded in IP address netid and hostid

Buffer Reassembly Problem Not all IP implementations are equally robust Some IP implementations may not reassemble datagram fragments correctly Solution: Configure IP software for DF = 1 -- Problem: DF flag may not be configurable by network manager Fragmentation may be required if IP datagram traverses networks with small MTU

IP Trailers: 4BSD UNIX Software derived from BSD 4.2 UNIX may use alternate IP encapsulation Done for efficient memory management -- To place data information on page boundary Berkeley-style trailer encapsulation Will not interoperate with normal IP encapsulation (example: IP routers)

Avoiding IP Trailers On many UNIX systems, IP trailer encapsulation can be controlled by the if conf ig utility Example: ifconfig ethO -trailers ifconfig ethO ethO: flags=23<UP,BROADCAST,NOTRAILERS> inet 144.19.74.201 netmask ffffOOOO broadcast 144.19.255.255

All O's Broadcast IP software implemented on earlier BSD 4.2 UNIX may use all O's broadcast Can cause confusion with most systems that use all 1's broadcast Use if conf ig utility to enable all 1's broadcast May not work if broadcast mechanism has been hard-coded Example: ifconfig ethO broadcast 144.19.255.255 ifconfig ethO ethO: flags=23<UP,BROADCAST,NOTRAILERS> inet 144.19.74.201 netmask ffffOOOO broadcast 144.19.255.255

The Internet Protocol

Using Unique Internet Addresses If building your own private internet Decide on an IP address class -- Popular choices are class B, class C If you decide not to connect to the Internet You can select your own IP network number If you decide to connect to the Internet You should apply to Network information Center (NIC) for unique internetwork number Alternatively, use IP address translation devices such as application-level gateways -- Application-level gateways also can be used to implement -- firewalls for enhancing security -- Can be used to avoid duplicate IP address conflicts

Application-Level Security Gateway (Firewall) Application-level firewall provides Isolation between duplicate IP addresses Security by restricting access between internal and external networks at the Application Layer Alternatively, use a private address

Obtaining Unique IP Network Number To obtain Internet number to connect to the Internet, apply to Network Solutions InterNIC Registration Services 505 Huntmar Park Drive Herndon, VA 22070 USA HOSTMASTER@INTERNIC.NET Users wanting to connect to the MILNET must still apply to DDN Network Information Center 14200 Park Meadow Drive, Suite 200 Chantilly, VA 22021 USA HOSTMASTER@NIC.DDN.MIL See Appendix D for application form for Obtaining IP Network Number

IP Address Database Local management of IP addresse RARP servers can Configuration database kept on local machine -- Often simple text files, such as /etc/hosts (UNIX) or net . cfg and config . tel, etc. (MS-DOS) RARP servers BOOTP server DHCP server (covered in Course 154) Name servers RARP servers can Simplify IP address maintenance Problems: -- Updates when data-link address changes -- RARP storms -- Single point of failure

The Internet Protocol

IP on IEEE 802 LANs Initial IP implementation on LANs was on Ethernet Ethertype field in Ethernet header is used to indicate IP packet -- Ethertype = 800 hex for IP packets There is no Ethertype field in IEEE LANs How do you indicate Ethertype information? -- Use lEEE802.2LLC

IEEE 802.2 Logical Link Control IEEE LANs use a sublayer called LLC to indicate protocol (software) addresses

The SNAP Protocol A special DSAP or SSAP value in IEEE 802.2 field indicates that Ethertype field is in the data field This mechanism is called SubNet Access Erotocol (SNAP)

IP on IEEE 802.3, 802.5, and FDDI

IP on X.25 First octet in X.25 call request data field indicates IP protocol X.25 virtual circuit is used to transmit datagrams -- Closed after a period of inactivity -- Treated as a point-to-point circuit

IP on ATM ATM transmits data in fixed 53-byte cells (5 bytes header and 48 bytes data) ATM cells available on demand with low latency -- Real-time audio/video -- Multimedia applications ATM provides virtual channels with quality of service (QoS) parameters

Chapter Summary You have learned about The role of the Internet Protocol (IP) IP address classes Address Resolution Protocols IP packet structure IP on various physical networks